Research On Auto-focusing Theory And Technology Based On Image Processing

Auto-focusing technology is a key technology in the robot vision and imaging system and various optical precision instruments.With the rapid development of science and technology,the auto-focusing problem has been much concerned by people,and its application has been more and more important.It can improve the accuracy and speed of measurement,automatic level and reduce the laboring intensity etc.From the late years of 80's,domestic and foreign scholars have made many researches and achievements in the auto-focusing field.However,at present there are still many shortcomings in auto-focusing techniques used in digital imaging systems because of the complexity of imaging objects and conditions.Such as how to select focus functions and focus windows,how to perform feedback control,and how to improve the auto-focusing velocity etc.,are far from completely solved.Firstly,much work of predecessors is summarized,and the development of the auto-focusing technique at home and abroad is introduced as well as the basic methods of auto-focusing.Based on Fourier optical theory,PSF(Point Spread Function)and OTF(Optical Transfer Function)are introduced.Then the mechanism why digital image is focused in optics imaging system is also analyzed.Then,the three key technologies in auto-focusing are analyzed.The focusing window directly affects the speed and accuracy of auto-focusing.Based on the importance of the focusing window,the necessity of windows selection is verified through analyzing a simple and functional optical imaging model,at last,the general strategy of windows selection in different imaging situation are brought forward.Aiming at the precision and anti-noise ability of the function,based on the comparisons and researches of some well-known auto-focusing functions and edge detection algorithms,a new clarity-evaluation function was advanced.In this function, an improved Sobel function was adopted,to some extent,the problem of the limited orientations in other clarity-evaluation functions was solved effectively.Meanwhile, standard deviation as a threshold parameter was introduced to reduce the influence of noise.The results show that this function has the characteristics of powerful single apex,high sensitive and strong anti-nosie ability.Peak searching algorithm realizes the searching and orientation of the focal point, through the comparison of the evaluation function values to decide the moving steps and directions,and get the feedback to control until gets to the best imaging position. Several commonly used peak searching algorithms are introduced,including Fibonacci Searching,Function Approximation Method,and Mountain Climbing Searching.Aiming at the limits of traditional climbing searching,an improved Mountain Climbing Searching is proposed.In the last chapter,hardware and software architecture are also briefly discussed. Also,the conclusion of my work and future expectation are proposed.